1. Field of the Invention
The present invention relates to a position detector, a position detection method, an exposure apparatus, and a method of manufacturing a device.
2. Description of the Related Art
A projection exposure apparatus which projects and transfers a pattern drawn on a reticle or photomask onto, for example, a wafer by a projection optical system has conventionally been employed to manufacture, for example, a semiconductor device, liquid crystal display device, or thin-film magnetic head by using photolithography. A projected image of the mask pattern formed via the projection optical system is aligned with a pattern, which has already been formed on the wafer, by an alignment detection system mounted in the projection exposure apparatus, and then exposure is performed.
Along with advance in micropatterning and an increase in packing density of integrated circuits, the projection exposure apparatus is required to project and transfer a mask pattern onto a wafer by exposure with a higher resolution. A minimum line width (resolution) that the projection exposure apparatus can transfer is proportional to the wavelength of light for use in exposure, and is inversely proportional to the numerical aperture (NA) of the projection optical system. According to this principle, the shorter the wavelength, the better the resolution. In view of this, the light source is currently shifting from the g-line (wavelength: about 436 nm) and the i-line (wavelength: about 365 nm) of a superhigh pressure mercury lamp to a KrF excimer laser (wavelength: about 248 nm) and an ArF excimer laser (wavelength: about 193 nm). The practical application of an F2 laser (wavelength: about 157 nm) to the light source is also in progress. Even EUV (Extreme Ultra-Violet) light having a wavelength of several to 100 nm is expected to be adopted in the future.
To further improve the resolution of the exposure apparatus, an immersion exposure apparatus has been put on the market, which increases the NA by filling at least part of the space between the projection optical system and the wafer with a liquid having a refractive index higher than 1. In the immersion exposure apparatus, the space between the wafer and an optical element which constitutes the end face of the projection optical system on its wafer side is filled with a liquid having a refractive index close to that of the photoresist layer. This makes it possible to increase the effective numerical aperture of the projection optical system seen from its wafer side, thus improving the resolution.
In this manner, along with the shortening of the wavelength of the exposure light, and the advent of the immersion method, the resolution is increasingly improving. To keep up with this trend, a higher overlay accuracy of the wafer is also required. In general, an overlay accuracy of about ⅕ the resolution is necessary. Hence, an improvement in overlay accuracy is increasingly becoming important for advance in micropatterning of semiconductor devices.
Roughly stated, two types of wafer alignment detection systems have been proposed and are in use already. The first system is a so-called off-axis alignment detection system (to be referred to as an OA detection system hereinafter) which is configured separately from the projection optical system and optically detects an alignment mark on the wafer. The second system is an alignment detection system which detects an alignment mark on the wafer using the alignment wavelength of non-exposure light through a projection optical system of the so-called TTL-AA (Through The Lens Auto Alignment) scheme used especially as the alignment scheme of an i-line exposure apparatus.
In the above-described alignment detection systems, the measurement result often has errors resulting from noise components of a detection signal, which is obtained by illuminating and observing the observation target surface, such as the distortion of the detection system due to, for example, an illumination variation, a variation in sensitivity of, for example, the light-receiving element or image pickup element, or dust adhesion on the detection system itself. Considering the current trend toward a higher resolution, it is important to reduce these measurement errors. To reduce these measurement errors, a process of removing these noise components from the detection signal (to be referred to as “noise correction” hereinafter) is performed. The noise correction is a process of measuring in advance various types of noise components of a detection signal obtained by illuminating and observing the observation target surface, storing them as noise signals, and correcting the detection signal of the alignment mark by referring to the noise signals.
The noise signals for noise correction are stored for each normal alignment condition (illumination wavelength, illumination NA, and detection NA), and the noise correction is performed using a noise signal corresponding to each alignment condition. Japanese Patent Laid-Open No. 11-54418 discloses details of such a prior art.
Conditions with regard to the alignment performance of the wafer alignment detection system mounted in the exposure apparatus often change due to a temporal change in the properties of an optical system in the detection system, so they are adjusted as needed. The conditions with regard to the alignment performance include, for example, the aberration, a shift of the optical axis, and a shift component for each wavelength (to be referred to as a “wavelength shift difference” hereinafter), which is generated due to decentration of, for example, a lens or plane-parallel plate. Assume that noise signals used before the adjustment have become no longer optimum although the alignment conditions remain the same. In this situation, as alignment is performed using the previous noise signals, measurement errors are likely to occur.